This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
STOMP is a randomized, parallel, open-label phase 1/2 study in multiple myeloma evaluating selinexor and backbone treatments across 12 arms. The registered primary endpoints span dose-escalation decisions, pharmacokinetic exposure, and phase 2 measures of tumor response and clinical benefit.
| Feature | STOMP |
|---|---|
| Trial acronym | STOMP |
| NCT identifier | NCT02343042 |
| Phase | Phase 1/2 |
| Condition | Multiple Myeloma |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 300 |
| Number of arms | 12 |
| Lead sponsor | Karyopharm Therapeutics Inc |
| Sponsor type | Industry |
| Status | Recruiting |
| Study start | 2015-10 |
| Primary completion date | 2027-04 |
2. Clinical Question
The central clinical question is how selinexor performs when evaluated with different backbone treatments in patients with multiple myeloma, including how its dose should be selected, how exposure changes with clarithromycin, and how each treatment arm performs with respect to objective response and clinical benefit.
Population
Patients with multiple myeloma enrolled in a randomized phase 1/2 treatment study.
Intervention
Selinexor is evaluated alongside backbone treatments in the registered study arms.
Backbone treatments
The registry identifies dexamethasone, lenalidomide, pomalidomide, bortezomib, daratumumab, carfilzomib, ixazomib, elotuzumab, clarithromycin, belantamab mafodotin, and mezigdomide among the study interventions.
Primary questions
Which doses are appropriate for further development, what are the selinexor pharmacokinetic exposure measures, and what are the response and clinical-benefit outcomes in phase 2 expansion?
3. Trial Design
Registered interventions
- Selinexor
- Dexamethasone
- Lenalidomide
- Pomalidomide
- Bortezomib
- Daratumumab
Additional backbone treatments
- Carfilzomib
- Ixazomib
- Elotuzumab
- Clarithromycin
- Belantamab Mafodotin
- Mezigdomide
The registry identifies 12 study arms but does not provide arm-specific enrollment counts in the record summarized here. It also does not specify a single control treatment applicable across all 12 arms. Consequently, the statistical comparison is best understood as a multi-arm development program rather than as one two-group treatment-versus-control comparison.
4. Study Timeline
STOMP begins
The registry records the study start in October 2015.
Dose and exposure characterization
The registered phase 1 endpoints evaluate maximum tolerated dose, recommended phase-2 dose, and selinexor pharmacokinetic measures.
Clinical activity assessment
The expansion phase evaluates overall response rate, duration of response, and clinical benefit rate independently for each arm.
Registered primary completion
The registry lists April 2027 as the primary completion date.
5. Endpoints
STOMP has a deliberately heterogeneous endpoint structure. The phase 1 endpoints address dose selection and pharmacokinetics, while the phase 2 expansion endpoints address antitumor activity. This distinction matters because these endpoints require different statistical estimands and analysis frameworks.
| Phase | Primary endpoint | Registry time frame | Registry definition |
|---|---|---|---|
| Phase 1 | Maximum Tolerated Dose (MTD) | 12 months | MTD for once weekly and twice weekly selinexor dose cohorts in the 11 Arms will be evaluated. |
| Phase 1 | Recommended Phase-2 dose (RP2D) | 12 months | RP2D for each Arm will be determined. |
| Phase 1 | Maximum Plasma Concentration (Cmax) of Selinexor | Pre-dose, 1 hour, 1.5, 2, 3, 4, 5, 6, 8, and 24 hours post-dose on Day 1 without clarithromycin and Day 8 with clarithromycin | Cmax of selinexor over a dosing interval when given with and without clarithromycin. |
| Phase 1 | Area Under the Concentration-time Curve From Time Zero to the Last Non-zero Concentration (AUC0-t) of Selinexor | Pre-dose, 1, 1.5, 2, 3, 4, 5, 6, 8, and 24 hours post-dose on Day 1 without clarithromycin and Day 8 with clarithromycin | Total exposure of selinexor in the blood (AUC0-last) from the time of dosing to the last measurable concentration collected when given with and without clarithromycin. |
| Phase 1 | Area Under the Concentration-time Curve From Time Zero to Infinity (Extrapolated) (AUC0-inf) | Pre-dose, 1, 1.5, 2, 3, 4, 5, 6, 8, and 24 hours post-dose on Day 1 without clarithromycin and Day 8 with clarithromycin | The registry specifies the endpoint and sampling schedule but does not provide an additional endpoint description. |
| Phase 2 | Overall response rate (ORR) | 12 months | ORR for each Arm independently. ORR to include stringent complete response (sCR), complete response (CR), very good partial response (VGPR), and partial response (PR), according to the International Myeloma Working Group (IMWG) criteria. |
| Phase 2 | Duration of response (DOR) | 12 months | Duration of response for each Arm. DOR is defined as the number of days from the date of the first evidence of objective response until progression. |
| Phase 2 | Clinical Benefit Rate (CBR) | 12 months | CBR is defined the point estimate of the percentage of patients in that arm who have a response of sCR, CR, VGPR, PR or Minimal response (MR), as assessed by IMWG criteria. |
6. Statistical Methodology
Phase 1 dose escalation
The phase 1 component is designed around dose finding rather than a conventional confirmatory comparison. The registered MTD endpoint asks for the maximum tolerated dose for once-weekly and twice-weekly selinexor dose cohorts in 11 arms, while the RP2D endpoint asks for the recommended phase-2 dose for each arm.
Statistically, dose escalation is different from testing whether one fixed dose is superior to another. The objective is to characterize the relationship between dose and tolerability and to identify a dose suitable for subsequent evaluation. The precise dose-escalation algorithm, toxicity probability target, cohort size, stopping rule, and decision boundary are not reported in the ClinicalTrials.gov record summarized here.
Recommended phase-2 dose
RP2D is a development decision that integrates information accumulated during dose escalation. It is not itself an effect estimate such as a hazard ratio or risk ratio. The statistical interpretation therefore depends on the decision rule used to move from observed tolerability and exposure data to the selected dose.
The important statistical point is that an RP2D is a selected development dose, not a population-average treatment effect. The uncertainty surrounding that selection depends on the underlying dose-escalation design and the number of observations available at each dose.
Pharmacokinetic analysis
The PK endpoints use serial selinexor concentrations collected from pre-dose through 24 hours after dosing. Cmax summarizes the maximum observed plasma concentration during the specified dosing interval. AUC0-t summarizes exposure from dosing through the last measurable concentration, while AUC0-inf extends exposure to infinity using extrapolation beyond the observed sampling period.
These measures describe drug exposure rather than clinical response. Their interpretation depends on the sampling schedule, concentration measurements, and the method used to estimate exposure between observed time points and beyond the last measurable concentration.
Phase 2 response analysis
ORR is defined as the percentage of patients in each arm with sCR, CR, VGPR, or PR according to IMWG criteria. Because each arm is evaluated independently, the natural primary summary is an arm-specific response proportion with an appropriate confidence interval.
CBR extends the response definition to include Minimal response (MR), so its numerator can include patients meeting sCR, CR, VGPR, PR, or MR according to IMWG criteria. The resulting percentage is an arm-specific point estimate rather than a time-to-event measure.
Duration-of-response analysis
DOR is defined as the number of days from the first evidence of objective response until progression. This makes DOR fundamentally different from ORR. ORR asks whether a response occurred; DOR asks how long the response persisted among patients who achieved an objective response.
For a DOR endpoint, patients who have not progressed by the relevant observation point may be censored rather than treated as if they experienced progression on the last observation date. Consequently, a time-to-event method such as Kaplan-Meier estimation is generally more appropriate than simply averaging observed response durations.
7. Planned Analysis
MTD and RP2D
The registry says that MTD will be evaluated for once-weekly and twice-weekly selinexor dose cohorts in 11 arms and that RP2D will be determined for each arm. For an endpoint of this type, the analysis would ordinarily summarize observed dose-limiting toxicity or tolerability information by dose and apply the prespecified dose-escalation decision rule to identify the dose considered appropriate for phase 2. The registry does not report that decision rule.
Cmax, AUC0-t, and AUC0-inf
For the PK endpoints, the registered sampling schedule provides repeated concentration measurements on Day 1 without clarithromycin and Day 8 with clarithromycin. Cmax would typically be summarized from the observed concentration-time profile, while AUC0-t and AUC0-inf would typically be calculated using a pharmacokinetic analysis method appropriate to the observed concentration data and terminal phase.
The with-versus-without-clarithromycin design makes exposure comparison an important component of the PK interpretation. A useful analysis would distinguish the within-person or between-condition exposure contrast, depending on the actual study design and analysis population specified in the statistical analysis plan. The ClinicalTrials.gov record does not provide that analysis specification.
ORR and CBR
For ORR and CBR, the registry specifies a point estimate of the percentage of patients in each arm meeting the relevant IMWG response definition. A standard analysis would report the response proportion for each arm together with a confidence interval reflecting the number of evaluable patients. Because there are multiple arms, the analysis should preserve the arm-specific nature of the endpoint rather than collapse the study into a single overall response rate.
DOR
DOR is defined from first objective response until progression. A conventional analysis would use a time-to-event framework, typically including Kaplan-Meier estimation when censoring is present. The resulting summary could include a median DOR and a confidence interval when sufficient events and follow-up are available. No such results are posted in the ClinicalTrials.gov record.
8. Statistical Methods Explained
Why are MTD and RP2D different?
MTD and RP2D answer related but distinct development questions. MTD concerns the upper dose identified through the dose-escalation process, whereas RP2D identifies the dose selected for subsequent phase-2 evaluation. The RP2D does not have to be interpreted simply as the numerical maximum dose observed in the study; its determination depends on the prespecified development decision process.
What does Cmax measure?
Cmax is the maximum plasma concentration observed over the specified dosing interval. It describes peak exposure. It does not measure total drug exposure, which is why AUC endpoints are also registered.
Why are both AUC0-t and AUC0-inf included?
AUC0-t uses the concentration measurements through the last non-zero measurable concentration. AUC0-inf adds an extrapolated component representing exposure after the last observed concentration. Reporting both allows the observed portion of exposure to be distinguished from the total exposure estimate that depends partly on extrapolation.
Why does the clarithromycin comparison matter statistically?
The registered PK measurements are collected on Day 1 without clarithromycin and Day 8 with clarithromycin. This creates a structured exposure comparison. The statistical interpretation depends on whether the relevant comparison is within participants or between distinct groups, as well as on the analysis population and pharmacokinetic model used. Those details are not specified in the registry record.
Why is ORR not a survival endpoint?
ORR is a proportion: a patient either meets the registered response definition or does not. There is no time component in the ORR definition itself. DOR is different because it begins at first objective response and continues until progression, creating a time-to-event structure.
Why would DOR typically use Kaplan-Meier analysis?
Patients may remain in response at the end of observation, so their exact future duration is unknown. Those observations can be right-censored. Kaplan-Meier methods are designed to estimate a time-to-event distribution while retaining information contributed before censoring.
Why should each arm be analyzed independently for ORR?
The registry explicitly defines ORR for each arm independently. With 12 registered arms, combining all participants would obscure differences between the individual treatment combinations and would no longer correspond to the registered endpoint definition.
9. Interpreting the Phase 1 Endpoints
MTD
A dose-selection endpoint focused on tolerability during dose escalation. It is not a measure of antitumor response.
RP2D
A development decision identifying the dose selected for phase-2 evaluation in each arm.
Cmax
A peak-concentration measure derived from the serial plasma concentration profile.
AUC
An exposure measure that integrates concentration over time. AUC0-t uses observed measurable exposure; AUC0-inf includes extrapolation to infinity.
These endpoints should not be treated as interchangeable evidence. A dose may generate a particular exposure profile without directly establishing clinical efficacy, while an observed response rate does not by itself determine whether the corresponding dose has an acceptable tolerability profile. The phase 1 and phase 2 components therefore answer complementary questions.
10. Interpreting the Phase 2 Endpoints
| Endpoint | What it measures | Statistical form |
|---|---|---|
| ORR | Percentage with sCR, CR, VGPR, or PR according to IMWG criteria | Arm-specific binary response proportion |
| DOR | Number of days from first evidence of objective response until progression | Time-to-event endpoint among responders |
| CBR | Percentage with sCR, CR, VGPR, PR, or MR according to IMWG criteria | Arm-specific binary clinical-benefit proportion |
The three endpoints provide different views of antitumor activity. ORR focuses on objective response, CBR broadens the response definition by adding Minimal response, and DOR asks how long an objective response persists. A complete interpretation therefore requires attention to the endpoint definition rather than treating all response-related percentages as equivalent.
11. Multiple Arms and Statistical Interpretation
The presence of 12 arms creates an important multiplicity issue. Each independently analyzed arm can generate its own estimate, confidence interval, and statistical test if formal testing is undertaken. As the number of comparisons increases, the probability of obtaining at least one apparently unusual result by chance also increases when conventional hypothesis tests are repeatedly applied without adjustment.
The appropriate multiplicity framework depends on which comparisons are considered confirmatory and which are descriptive or exploratory. The ClinicalTrials.gov record does not specify a multiplicity-adjustment procedure for the 12-arm program.
This does not make arm-specific estimates unusable. It means that the statistical meaning of a nominal p-value depends on the comparison family and prespecified testing strategy. For an early-phase multi-arm development study, descriptive estimation can be more informative than treating every arm-level comparison as an independent confirmatory hypothesis test.
12. Randomization and the Parallel-Arm Structure
Randomization is a key design feature because, when successfully implemented, it creates the basis for comparing treatment groups without relying on measured baseline characteristics to establish comparability. STOMP is registered as randomized and parallel, with no masking.
The parallel structure means that participants are assigned to study arms that are evaluated concurrently rather than being repeatedly switched between treatment conditions. The registry does not identify a crossover design, and no crossover endpoint or crossover analysis is specified in the registered primary endpoints.
13. Pharmacokinetic Sampling and Analysis
The PK endpoints use a detailed sampling schedule extending from pre-dose through 24 hours after dosing. The same general sampling times are registered for Day 1 without clarithromycin and Day 8 with clarithromycin.
| PK endpoint | Sampling schedule | Statistical interpretation |
|---|---|---|
| Cmax | Pre-dose, 1, 1.5, 2, 3, 4, 5, 6, 8, and 24 hours | Peak plasma concentration over the specified dosing interval. |
| AUC0-t | Pre-dose, 1, 1.5, 2, 3, 4, 5, 6, 8, and 24 hours | Observed exposure from time zero through the last measurable concentration. |
| AUC0-inf | Pre-dose, 1, 1.5, 2, 3, 4, 5, 6, 8, and 24 hours | Total exposure estimate including extrapolation beyond the observed concentration period. |
Sampling density matters because PK parameters are derived from the concentration-time profile rather than from a single observation. Missing or poorly timed samples can affect the precision of Cmax and AUC estimates. For AUC0-inf in particular, the extrapolated terminal component can introduce additional uncertainty because the portion after the last measurable concentration is not directly observed.
14. Response Endpoints and Estimation
ORR and CBR are naturally expressed as proportions. If an arm has \(r\) responders among \(n\) evaluable patients, the point estimate is \(r/n\). The statistical uncertainty around that proportion should be represented with an appropriate binomial confidence interval.
The confidence interval describes uncertainty in the estimated arm-level response proportion. It does not describe the probability that an individual patient will respond.
Because STOMP contains multiple arms, the denominator is particularly important. A response percentage is meaningful only when its analysis population is clear. The registry identifies the endpoint as applying independently to each arm, but the record summarized here does not provide arm-specific sample sizes or results.
15. Duration of Response as a Time-to-Event Endpoint
DOR begins only after a patient has demonstrated the first evidence of objective response. This creates a selected analysis population: patients who never respond cannot contribute a DOR duration under the registered definition.
A long observed DOR is not equivalent to a high ORR. ORR measures how frequently an objective response occurs, whereas DOR measures persistence after that response has occurred. An arm could therefore have a lower response proportion but a long duration among its responders, or a higher response proportion with shorter observed response durations.
For DOR, censoring rules are important. A patient who remains in response when follow-up ends does not necessarily have a known complete duration. A time-to-event analysis can retain the observed response time while treating the unresolved future duration as censored according to the prespecified rules.
16. Limitations
- No posted statistical analyses: the ClinicalTrials.gov record contains no posted statistical-analysis outputs or numerical endpoint results for the registered primary endpoints.
- Arm-specific enrollment is not reported here: although total planned enrollment is 300 and the study has 12 arms, the registry data summarized here do not provide the enrollment assigned to each arm.
- Comparator structure is not fully specified: the registry lists the interventions and 12-arm parallel randomized design but does not provide a single common comparator applicable to the complete study.
- Dose-escalation algorithm is unspecified: the registry identifies MTD and RP2D but does not provide the exact dose-escalation rule, toxicity probability target, or decision boundaries.
- PK analysis details are unspecified: the sampling schedule is registered, but the record does not state the precise pharmacokinetic estimation method, handling of missing samples, or statistical comparison model.
- Multiplicity strategy is unspecified: the presence of 12 arms creates a potentially large comparison family, but the registry does not state a multiplicity-adjustment procedure for the primary endpoints.
- Analysis populations are not fully described: the registry defines the endpoints but does not provide detailed statistical-analysis populations for every endpoint.
- Ongoing study: the registry lists the status as Recruiting and the primary completion date as 2027-04, so the study remains within its registered development period.
17. Why This Trial Matters Statistically
STOMP is a useful statistical teaching case because it combines several distinct classes of clinical-trial endpoint within a randomized multi-arm development program. The study is not simply a conventional two-group efficacy trial. It links dose selection, pharmacokinetics, response assessment, and duration of response in a single development framework.
| Concept | How it appears in STOMP |
|---|---|
| Randomization | Participants are assigned randomly within a 12-arm study. |
| Parallel-arm design | The registry identifies a parallel design rather than a crossover design. |
| Phase 1 dose escalation | MTD and RP2D are registered primary endpoints. |
| Pharmacokinetics | Cmax, AUC0-t, and AUC0-inf are evaluated using serial plasma sampling. |
| Exposure comparison | PK sampling is registered both without and with clarithromycin. |
| Binary response analysis | ORR and CBR are defined as arm-specific percentages. |
| Time-to-event analysis | DOR measures time from first objective response until progression. |
| Multiple-arm inference | 12 arms create a broader comparison structure than a simple two-arm trial. |
| Endpoint heterogeneity | Dose, PK, response, and duration endpoints require different statistical frameworks. |
| Open-label design | The registered masking status is none. |
18. Statistical Concepts in This Trial
STOMP illustrates why clinical-trial analysis begins with the estimand and endpoint definition rather than with a generic statistical test. MTD and RP2D concern dose selection; Cmax and AUC describe pharmacokinetic exposure; ORR and CBR describe the proportion of patients meeting response criteria; and DOR describes the persistence of response over time.
The most important statistical discipline is therefore matching the analysis to the endpoint. A single method applied indiscriminately across all endpoints would fail to reflect the underlying data structure.
19. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The registered endpoints represent distinct statistical estimands. Dose-selection endpoints require decision rules, PK endpoints require concentration-time analysis, response endpoints require proportion estimation, and DOR requires time-to-event methods when censoring is present.
Clinical interpretation
The study is structured to examine selinexor across multiple backbone-treatment settings, first establishing dose and exposure characteristics and then evaluating response and clinical benefit within the phase 2 expansion arms.
Because no formal statistical analyses have been posted to ClinicalTrials.gov, the registered endpoints define the questions that the study is designed to answer, while the numerical magnitude and uncertainty of those outcomes are not available in the registry record summarized here.
20. Record Summary
STOMP is a randomized, parallel, open-label phase 1/2 study in multiple myeloma with 12 registered arms and planned enrollment of 300 participants. Its primary endpoint structure is unusually broad: phase 1 evaluates MTD, RP2D, Cmax, AUC0-t, and AUC0-inf, while phase 2 evaluates ORR, DOR, and CBR independently for each arm. This makes STOMP a useful example of how dose-selection, pharmacokinetic, binary-response, and time-to-event methods fit together within a single clinical development program.
Continue through the Clinical Biostats clinical-trial library
Clinical trial analysis connects study design, endpoint definitions, statistical methods, and interpretation into a single framework for understanding clinical evidence.
21. Sources
- ClinicalTrials.gov: NCT02343042 — STOMP: Selinexor and Backbone Treatments of Multiple Myeloma Patients.
- PubMed: PubMed record.
- PubMed: PubMed record.
- PubMed: PubMed record.